After the close Xilinx (XLNX) reported better-than-expected results, but guidance fell short and the stock drifted lower. I think this could be a good opportunity to add to a high quality semiconductor name.
I was cautiously optimistic back in January 2012 on a handful of semiconductor names. At the time, I thought we were at the start of a new semiconductor cycle driven by end market demand and die shrinks. I've been following the semiconductor business since the early 1990s and I've seen quite a few of these cycles.
According to Infonetics and Xilinx, spending on wireless communications like LTE and optical networks are growing like crazy. For example, capital spending on 4G LTE is expected to grow from $6.5 billion in 2012 to an estimated $17 billion by 2016. That works out to a consolidated annualized growth rate of 27%.
Optical networking gear is growing as all those wireless users consume serious amounts of bandwidth. Spending on optical network build outs are expected to grow to $14.8 billion by 2016 from $8.2 billion in 2012. Other markets like switching, routing and data networking will continue to churn out steady mid single digit growth.
Xilinx is a big player in these markets, since its field programmable gate array (FPGA) semiconductors give equipment makers a leg up on the competition by speeding up the design process. FPGA chips can be programmed and updated to provide additional functionality. You can drop an FPGA into your design and program the features you need.
Other semiconductors, such as application-specific integrated circuits (ASICs), are custom designed for a single purpose. ASICs are great for devices where the design doesn't change for years, and the chips address a large end market. A motherboard in a personal computer usually contains a lot of ASICs. It can take 2.5 years to engineer, test and tape out a new ASIC design. That's too slow for the wireless or optical transport industry.
FPGA are used in low volume, high margin equipment like optical networking gear. In order to justify a 28 nanometer (nm) ASIC, you would have to sell something like $600 million worth of chips. That's a lot of chips to move. Even if you found a market large enough to absorb that many chips, the next time you need to shrink the design, you would need to spend between 5 and 10 times your revenue to get from a 20 nm design to a 16 nm design. The smaller the die size, the more research and development dollars you consume, and that forces you to find larger end markets. But with an FPGA, you can address smaller markets. In that same example, you'd need to sell about $135 million worth of FPGAs, instead of $600 million of ASICs.
Because of that inherent advantage, FPGAs sell at a premium to other semiconductors. Companies like XLNX and its main competitor, Altera (ALTR), generally have margins in the mid-to-high 60s, about 10-15 points higher than other semi makers. According to Xilinx, 40% of the company's design wins have displaced ASIC designs from other vendors.
Market share gains, high growth end markets and better margins will keep Xilinx moving higher over the next three years.
For fiscal year 2014, the Street expects XLNX to grow between 8-11%. If the company were to achieve the high end of expectations, XLNX would produce about $2.4 billion in revenue. Gross margins are expected to tack on an additional 300 basis points, which works out to a 69% gross margin. Because of the earnings leverage, XLNX could increase earnings per share by as much as 24%.
After the close, the company guided the third quarter down, but that's not unusual. The semiconductor business can be lumpy. New products get delayed. You can divide a single fiber strand into as many as 160 channels and pump down 1.6 terabits at 97% the speed of light. So, you can forgive the companies when they miss a quarter or two. It's not exactly easy to get this stuff working. Typically, the companies will make up the shortfall later on, because the demand for wireless and optical networking isn't going away.
If the stock sold off on Thursday, I think it would be an excellent time to build a position. These cycles usually last three years, so you have time to pick your spot.